Re: general protection fault in drm_atomic_set_crtc_for_connector

From: Desmond Cheong Zhi Xi <hidden>
Date: 2021-08-18 09:35:54

#syz fix: drm: fix oops in drm_atomic_set_crtc_for_connector

At the time of the bug report [1], there was the following call in 
update_output_state:
			ret = drm_atomic_set_crtc_for_connector(new_conn_state,
								NULL);
This was followed by
	if (crtc) {
		crtc_state = drm_atomic_get_crtc_state(conn_state->state, crtc);
		if (IS_ERR(crtc_state))
			return PTR_ERR(crtc_state);

		crtc_state->connector_mask |=
			drm_connector_mask(conn_state->connector);

		drm_connector_get(conn_state->connector);
		conn_state->crtc = crtc;

		drm_dbg_atomic(crtc->dev,
			       "Link [CONNECTOR:%d:%s] state %p to [CRTC:%d:%s]\n",
			       connector->base.id, connector->name,
			       conn_state, crtc->base.id, crtc->name);
	} else {
		drm_dbg_atomic(crtc->dev,
			       "Link [CONNECTOR:%d:%s] state %p to [NOCRTC]\n",
			       connector->base.id, connector->name,
			       conn_state);
	}
in drm_atomic_set_crtc_for_connector, which clearly dereferences a null 
pointer in the else block. This led to the reported general protection 
fault.

commit 0003b687ee6d("drm: fix oops in 
drm_atomic_set_crtc_for_connector") fixes this by getting the device 
from connector, which can't be NULL.

Link: 
https://syzkaller.appspot.com/bug?id=768a24e51bd111e0c1d6b6f4e1f09fac3c54c05d 
[1]

Best wishes,
Desmond
_______________________________________________
Linux-kernel-mentees mailing list
Linux-kernel-mentees@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help